Jerric Chan

AI Graph Network Planner

Engineered an interactive, visual note-taking and flowchart application using React Flow to map complex relational data like legal cases. Integrated a vector database and generative AI recommendation engine to dynamically analyze chart states and surface relevant legal precedents. **Won 1st Place** at Nugen-Antler Hackathon. React, React Flow, Generative AI, Data Architecture.

AI Training Bot for Social Workers

Architected an interactive AI training chatbot simulator designed to bridge experiential gaps for social workers and crisis responders. Engineered a modular component layer for calibrating client personalities and difficulty levels alongside an adaptive emotional AI engine that simulates trust dynamics, behavioral resistance, and real-time state modeling. Integrated a post-session coaching dashboard leveraging Gemini to automatically evaluate chat transcripts, producing structured qualitative feedback, performance ratings, and actionable recommendations. Includes a full authoring workshop for supervisors to create custom clinical drills.

Smart-City Feedback & Analytics Dashboard

Built a production-ready business analytics data pipeline and interactive proof-of-concept dashboard to handle Customer Information System feedback data. Programmed machine learning components to automatically categorize department cases, assign topic labels, and dynamically register emerging trends. Integrated a custom semantic similarity module leveraging Natural Language Processing (NLP) to detect templated responses and flag coordinated public writing campaigns.

Derna Post-Disaster Flood Mitigation Analysis

Co-developed a scalable, post-disaster flood impact assessment and urban optimization model for Derna, Libya, following a catastrophic dam failure. Applied Object-Based Image Analysis (OBIA) on WorldView satellite remote sensing imagery and integrated the Google Places API to build automated building classification frameworks. Programmed an agent-based navigation routing system in Python to simulate vehicle versus pedestrian traffic through highly constrained, flooded corridors. Quantified cross-system destruction by mapping and classifying **3,680 damaged structures** (including cultural and economic hubs) and **27 km of ruined roadways**, delivering data-driven spatial insights to prioritize healthcare facility recovery.

Building & Scaling a DTC Personal-Care Brand

Built Dare Essence from 0 to 1 — iterating product formulations and scent profiles, then launching a custom Shopify storefront in React. Sourced and negotiated directly with manufacturers in China; when a production run came in out-of-spec, escalated to the supplier's leadership and secured a compensatory order of up to 20,000 units at a ~28% discount. Secured B2B consignment placements at boutique retail and fitness facilities. On growth, ran paid Meta Ads campaigns across two product launches — "Forging His Inner Sanctum" and "A Kindling of Amber" — pivoting from static product imagery to emotionally resonant video hooks after reading early ad metrics, moving the funnel from broad awareness to targeted seasonal conversions.

Verified Work Ledger

Chrome extension + web app that captures work in-context — from Jira, GitHub, and Slack — into structured ECO logs (Event, Contribution, Outcome), building a verified ledger of real impact instead of a resume rebuilt from memory months later. Site adapters pre-fill fields from ticket/PR state (a merged GitHub PR auto-suggests "Shipped to main"); a vector-search layer (FAISS, BERTopic) auto-scores and summarizes logs, and one click generates a Manager Calibration Brief or Peer Evaluation Brief ready to share. Designed the ECO framework itself, ran multi-tier user testing, and built and iterated it solo over 312 commits across 6 months.